INDEX

Explanations

instances where a brief view or insight is provided

oai_token-act-pair · gpt-3.5-turbo

phrases indicating insights or observations

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ransom

-0.69

die

-0.68

 Bulgar

-0.67

 Cosmos

-0.63

ylum

-0.59

 sacrific

-0.59

yne

-0.58

 dumped

-0.56

iety

-0.56

umb

-0.56

POSITIVE LOGITS

 glimpse

0.97

 glimps

0.94

 thereof

0.92

 impression

0.84

 largeDownload

0.78

 hint

0.76

 impressions

0.74

seeing

0.73

 hints

0.71

ittal

0.70

Activations Density 0.136%

instances where a brief view or insight is provided

phrases indicating insights or observations

No Comments

No Known Activations